User:Altoids0/Call them LLMs
From Wikipedia, the free encyclopedia
When referring to the text-generating chatbots which have gained notoriety and proliferation in the last several years, please strongly prefer using the neutral term LLM instead of the promotional and informal term "AI".
 | This is an essay on the terminology for LLMs and generative technologies indirectly promoted by essays and policies such as Wikipedia:Large language models. It contains the advice or opinions of one or more Wikipedia contributors. This page is not an encyclopedia article or a Wikipedia policy, as it has not been reviewed by the community. |
.jpg)
Call them LLMs
While referring to them as "AI" is indeed commonplace in journalism, scholarly sources tend to prefer referring to generative tools by the underlying technology,[1][2][3] meaning in technical contexts, or in contexts where the term "AI" is not used by your sources, it's going to be better to call them LLMs.
By referring to speculative generation tools as "AI" or especially "ChatGPTs", Wikipedians unfortunately collude with the marketing of these things by referring to them with such a high-prestige term. People come to this site every day and in good faith make use of LLMs on the understanding that they are intelligent and potentially smarter than them, when they are not. The language we use on the site should reflect the fact that we address these things as tools, and agree with the scholarly (and Wikipedian!) consensus that these things are generally unreliable when not deeply scrutinized.
Don't call them LLMs
Wikipedia is largely composed of text, so it is chiefly LLMs that affect this website specifically. Additionally, so far, LLMs are the only effective technology for producing text that sounds even remotely plausible beyond simple sentences. One would not get very far trying to generate a Wikipedia article using Markov chains.
That said, there are occasionally vandals who refuse to follow the guidelines concerning "enhanced" or outright fabricated images, or there may someday be new technologies obsolescing LLMs. You should not use the term "LLM" to refer to the tools used in those cases. This is unilaterally true for images, video and audio, as the technology used to generate them is distinct and changes more often. For example, tools like Stable Diffusion use a latent diffusion model, or LDM, to create bogus images. In the 2010s, generative adversarial networks (GANs) were more common, and still are occasionally used.
Lastly, despite everything said here, there are many people, including those who create reliable sources, who still refer to LLMs as AI (usurping video game NPCs, Clippy, phone keyboards apps with autocomplete and GLaDOS from the term), perhaps because LLMs are newer. Regardless, Wikipedians are allowed to prefer the terminology of their sources when appropriate, especially as Wikipedia is ultimately not a platform for social change. Do not link this essay at people who are simply quoting or acquiring knowledge from reliable sources who happen to prefer the informal terminology.